Effects of Groundwater Pumping
Content
return to menu
Effects Of Groundwater Pumping On Saltwater Intrusion
In The Lower Burdekin Delta, North Queensland
K. A. Narayana, C. Schleebergerb, P. B. Charleswortha and K. L. Bristowa
a
CSIRO Land and Water, Davies Laboratory, Townsville, Australia
b
University of Applied Sciences, Höxter, Germany
Abstract: The Burdekin Delta is situated in the dry tropics of North Queensland and provides a major water
resource for the irrigation of sugarcane. The Delta is unique in that it overlies a shallow groundwater system
and is close to the Great Barrier Reef. Aquifer management practices include large recharge pits to assist with
artificial replenishment of groundwater. Artificial recharge can be used to maintain groundwater levels and
subsequently control seawater intrusion. This technique, however, is often costly and ineffective in areas
where excessive groundwater pumping occurs. In the lower Burdekin Delta more than 1800 groundwater
pumps are used for irrigation purposes and excessive pumping has resulted in seawater intrusion near the
coastline. In this paper we describe the use of a variable density flow and solute transport model, SUTRA, to
define the current and potential extent of saltwater intrusion in the Burdekin Delta aquifer under various
pumping and recharge conditions. A 2-D vertical cross-section model has been developed for the area, which
accounts for groundwater pumping and various artificial recharge sites currently being used in the delta. The
Burdekin Delta aquifer consists mainly of sand and clay with granitic bedrock. The model domain uses
vertical cross sections along the direction of groundwater flow. The initial conditions were based on the land
use prior to agricultural development in the area when the seawater interface was in its assumed natural state.
The results address the effects of variations in pumping and artificial and natural recharge rates on the
dynamics of saltwater intrusion. The simulation has been carried out for a range of recharge, pumping rates
and hydraulic conductivity values. The results show that the saltwater intrusion is far more sensitive to
pumping rates and recharge than aquifer properties such as hydraulic conductivity. The impacts of possible
management scenarios on groundwater quality have also been investigated.
Keywords: Groundwater pumping, saltwater intrusion, SUTRA model, artificial recharge, Burdekin Delta
1.
Sorensen, 2001) and the first step in dealing with
these problems is to evaluate the size and extent
of the intrusion. The extent of this intrusion
depends on climatic conditions, the characteristics
of the groundwater flow within these aquifers,
and the manner of groundwater usage.
INTRODUCTION
The coastal areas of the world accommodate high
population with about 50% of the world
population lives within 60 km of the shoreline.
Overexploitation of the groundwater has become
a common issue along the coast where good
quality groundwater is available. Consequently,
many coastal regions in the world experience
extensive saltwater intrusion in aquifers resulting
in severe deterioration of the quality of
groundwater resources.
Saltwater intrusion problems in coastal aquifers
are not new. The initial model was developed
independently by Ghyben in 1888, and by
Herzberg in 1901. This simple model is known as
the Ghyben-Herzberg model and is based on the
hydrostatic balance between fresh and saline
water in a U-shaped tube. They showed that the
saltwater occurs at a depth h below sea level
represented by:
In Australia, coastal Queensland is fortunate to
have extensive groundwater resources. Many
rivers have well-developed alluvial tracts and
deltas with extensive sand and gravel aquifers.
The river delta systems usually contain rich soil
and were an obvious target for the development of
sugarcane plantations in the late 19th century.
Groundwater use for irrigation commenced
shortly after settlement, but it was not until the
expansion of the sugar industry in the mid-20th
century that irrigation was practiced extensively.
Serious problems of saltwater intrusion exist in
many coastal areas of Queensland (Volker and
Rushton, 1982; Hillier, 1993; Murphy and
h = >Us (Us-Uf)@ hf
(1)
where, Uf and Us are respectively the density of
fresh and saline water, and hf is the elevation of
fresh water level above mean sea level. Equation
1 is referred to as Ghyben-Herzberg relationship.
Substitution of Uf (1000 kg m-3) and Us (1025 kg
m-3) in eqn. (1) shows that h = 40 hf. In other
words, the depth to the fresh-saline interface
below mean sea level (h) is 40 times the elevation
224
return to menu
sugarcane areas in Australia. The topography is
essentially flat to slightly undulating, however,
outcrops of basement rock occur in the south and
southwest of the delta. Annual rainfall for the
region averages 1032 mm with most falling in the
December – March ‘wet’ season. Mean annual
evaporation is 2062 mm. The delta is situated in
close proximity to environmentally sensitive
wetlands, waterways, estuaries, and the Great
Barrier Reef. The location of the delta with regard
to its regional setting is shown in Figure 1.
of the water table above sea level (hf) (Freeze and
Cherry, 1979).
This simplistic model ignores convection,
dispersion and diffusion phenomena responsible
for the distribution of salinity in coastal aquifers.
In coastal aquifers, freshwater usually overlies the
seawater separated by a transition zone.
Management of limited groundwater resources in
such situations is a delicate task and requires
special attention to minimise the movement of the
saltwater wedge into aquifers and upconing of
saltwater near pumping stations. The extent of
intrusion depends on a number of factors such as
aquifer geometry and properties (hydraulic
conductivity,
anisotropy,
porosity
and
dispersivity), abstraction rates, and depth,
recharge rate, and distance of pumping wells from
the coastline (Ghassemi et al, 1993). Complex
models are required to quantify these factors.
The history of groundwater irrigation, aquifer
overdraft and consequent formation of the
Burdekin Delta recharge scheme is summarised
by Charlesworth et al. (2002). One of the main
purposes of the lower Burdekin artificial recharge
scheme is to maintain high water table levels to
keep the saltwater wedge from moving inland and
displacing the freshwater. Currently there are
more than 1800 groundwater pumps used for
irrigation water supply in the delta area. It is
critical to understand the effectiveness of the
artificial recharge scheme to withstand the
movement of saltwater wedge under various
irrigation and pumping regimes.
Over the years, several mathematical and
numerical models have been developed, which
serve to predict the interface or transition zone
between fresh groundwater of meteoric origin and
seawater in the subsurface of coastal areas (Reilly
and Goodman, 1985). The development of these
models was largely motivated by groundwater
issues; that is, assessment of fresh groundwater
reserves, and prediction of saltwater intrusion
(onshore salinity distribution) – the landward or
upward movement of the interface in response to
groundwater exploitation practices (e.g. Volker
and Rushton, 1982; Custodio et al., 1987;
Ghassemi et al., 1990; Ghassemi et al., 1993;
Gotovac et al., 2001).
In this paper we describe the use of a variable
density flow and solute transport model, SUTRA
(Voss, 1984), to define the current and potential
extent of saltwater intrusion in the Burdekin Delta
aquifer under various pumping and recharge
conditions. A 2-D vertical cross-section model
has been developed for the area, which accounts
for groundwater pumping and various artificial
recharge schemes currently being used in the
delta. Model results are compared with limited
available data in the area. The impacts of possible
management scenarios on groundwater quality
have also been investigated.
2.
Figure 1. Location of the Lower Burdekin Delta
(BRIA - Burdekin River Irrigation Area, NBWB
– North Burdekin Water Board and SBWB –
South Burdekin Water Board)
A single-layer approach to modelling the
Burdekin Delta aquifers allows all minor
variations to be incorporated into a single
hydrostratigraphic unit (Arunakumaren et al.,
2000). Groundwater within the delta is
constrained by the unconsolidated alluvial and
deltaic sediments, which were deposited by the
Burdekin River and its distributaries. Typical
cross-sections of the delta sediments show that
strata layers usually are laterally discontinuous,
making cross-bore correlations difficult to
identify. Nonetheless, it is apparent from bore
hydrographs that there is a significant amount of
vertical hydraulic connection between sandy
units.
SITE DESCRIPTION AND GENERAL
HYDRGEOLOGY
The Burdekin Delta is a major irrigation area in
the dry tropics of North Queensland, with about
40,000 ha of irrigated crops. The environmental
and climatic conditions are ideal for the
production of sugarcane and the region has a
reputation of being one of the highest yielding
225
return to menu
require adequate discretisation in space and time
and appropriate initial and boundary conditions.
In this work the delta is being simulated as a
single-layer unconfined aquifer, with a bottom
slope defined by the position of the granitic
bedrock. The hydraulic conductivities of the
aquifer have been estimated based on aquifer
materials (Freeze and Cherry, 1979). Ongoing
improvements in hydraulic characterization will
be made as more data becomes available. A
typical cross section of the delta aquifer is shown
in Figure 2.
3.
3.2 SUTRA model
SUTRA (Voss, 1984), in conjunction with the
Argus-One Graphic User Interface, was chosen as
the basis for numerical modelling because of its
ability to solve density dependent groundwater
flow and variably saturated flow, and also
because it is readily available in source code
form. This model implements a hybridisation of
finite element and integrated finite difference
methods employed in the framework of a method
of weighted residuals. The hybrid method is the
simplest and most economical approach, which
preserves the mathematical elegance and
geometric flexibility of finite element simulation,
while taking advantage of finite difference
efficiency.
NUMERICAL MODELLING OF
SALTWATER INTRUSION
3.1 Recharge and groundwater extraction
Recharge to the groundwater system is by a
number of different processes. These include
infiltration of rainfall, artificial recharge through
pits and channels, river recharge, flooding, and
irrigation return flows. From an assessment of
these mechanisms, simulated annual groundwater
recharge between 1981 and 1997 varied between
330,000
and
650,000
ML
per
year
(Arunakumaren et al., 2000). For analyses in this
study, total maximum recharge rate was set at
1,250 mm/yr (490,000 ML/yr.
3.3 Discretisation, Boundary Conditions and
Model Parameters
To simulate saltwater intrusion in the Burdekin
Delta aquifer, a cross-sectional vertical slice with
a length of 5000 m, a depth of 30 m on the left
boundary, 45 m on the right boundary (facing
saltwater), and a thickness of 1 m was taken. The
model area was discretised to 7500 (500 x 15)
quadrilateral elements and 8016 (501 x 16) nodes.
The details of the model domain with boundary
conditions are shown in Figure 3.
Simulation of the saltwater intrusion requires a
series of assumptions and estimations to be made
about the various parameters involved. There are
no volumetric metered records for groundwater
pumping rates in the Burdekin Delta. The most
recent estimate indicates a volume of 210,000530,000 ML/yr is being extracted (Arunakumaren
et al., 2000; SKM, 1996).
Figure 2. A cross-section of the Burdekin Delta in
groundwater flow direction showing sediment
distribution (after Arunakumaren et al. 2000)
The physical problem along the coast in the Delta
is one of density-dependent groundwater flow and
saltwater intrusion (Voss, 1984; Narayan and
Armstrong, 1995). Prediction of the saltwater
wedge can be obtained by the solution of two
spatial differential equations describing the
‘conservation of mass of fluid’ and ‘conservation
of mass of salt’ in a porous medium. Numerical
procedures for solving the governing equations
226
Figure 3. Model domain and boundary
conditions
A steady state simulation with 200 mm natural
recharge per year was used to generate the
situation prior to agricultural development. The
result is a saltwater wedge that extends about 50
m inland. Concentration and pressure outputs
from steady state SUTRA runs were used as an
initial condition to simulate realistic short-time
effects of agricultural land use (groundwater
discharge and recharge) on the groundwater
system.
return to menu
The boundary conditions for the transport
simulation are dependent on its flow boundary
conditions. On the right hand model boundary a
hydrostatic pressure was imposed where pressure
is zero at the sea surface, and increases linearly
with depth. Based on field observations, a
groundwater head of 2 m was chosen at a distance
of 5km from the right hand model boundary. The
concentration (TDS) of the recharge water is
assumed to be 300 mg/L owing to irrigation with
river- and groundwater.
Longitudinal dispersivity
2.5 m
Transverse dispersivity
0.5 m
Molecular diffusivity of solute in
water, Dm
10-9 m²/s
Porosity, İ
0.3
Viscosity of water, P
0.001 kg·m/s
Groundwater head (left of model
boundary)
2m
Recharge range
500 – 1250 mm/yr
The hydraulic conductivity varies across the delta
from 10 to >300 m/d at the northern coast. In the
area chosen for the SUTRA model (see Figure 2)
K is estimated to be 10 to 100 m/d (Freeze and
Cherry, 1979), salinity of groundwater in the
cross-section as 500 mg/L (entire Burdekin Delta
aquifer ranges: 300 to 2000 mg/L). Transverse
dispersivity is even less well quantified than
longitudinal dispersivity from field observations.
The dispersivity values listed in Table 1 are an
estimate only, governed to some extent by the
practical limitations imposed by mesh size.
Distance between wells, Lp
400 m
Pumping rate, Qp
5 – 50 L/s
4.
One of the main purposes of the lower Burdekin
artificial recharge scheme is to maintain high
water table levels to keep the saltwater wedge
from moving inland and displacing the
freshwater. It is important to understand the
effects of varying recharge and groundwater
pumping on the movement of the saltwater
wedge. Based on the work of Arunakumaren et al.
(2000), simulation runs were conducted with
recharge values of 600, 900 and 1250 mm/yr. The
actual recharge differs seasonally during a year.
SUTRA cannot simulate the seasonal variations
properly, so the constant source value for
recharge is spread over one year (e.g. 1250 mm/yr
§ 0.0000396 kg/(m²·s)).
A total of 2187 pump locations were surveyed
across the delta by GPS, including 367 open
water pumps and 1811 groundwater pumps by
NR&M (Arunakumaren et al., 2000). In this
model, 12 perforated wells, with discharge rates
of 5, 10, 15, 20, 25 or 50 L/s and a depth of about
10 m have been used along the transect. Nodes
with negative fluid source describe these wells.
Average calculated distance between pumps Lp
[m] in the delta is about 470 m; for modelling
purposes we used Lp = 400 m. The groundwater
pumping rates vary across the delta because of
different irrigation practices, field size, and
precipitation as well as surface water availability.
To obtain the linear discharge values for a onemeter thick vertical slice, the pumping rates were
divided by Lp. For example, 10 L/s pumping rate
can be converted to a line sink (Qp') by 10/400
(kg/m·s) = 0.025 kg/(m·s). Model input
parameters are given in Table 1.
The pumping rates used in the model are 5, 10,
15, 20, 25 and 50 L/s. This corresponds to 0.4324.32 ML/day. The first four values are more
realistic for the Burdekin Delta. A series of salt
concentration contours for various recharge rates
(600-1250 mm/yr), hydraulic conductivities (10 100 m/d) and pumping rates (5–50 L/s) were
obtained. Saltwater concentration profiles for
1250 mm/yr recharge rate and a typical pumping
rate of 10 L/s are shown in Figure 4 after 10
years. Concentration profiles are influenced by
the left hand boundary for longer pumping
durations.
Table 1. Input parameters.
Value
Seawater concentration, CS
35,700 mg/L
Groundwater concentration, CG
500 mg/L
Recharge water concentration, CR
300 mg/L
Freshwater density, ȡF
1,000 kg/m³
Density change, wȡ/wC
700 kg²/(kgTDS·m³)
Hydraulic conductivity, K
10, 50, and 100 m/day
Metres
Parameter
RESULTS AND DISCUSSION
Figure 4. Saltwater concentration profile in a
vertical cross-section after 10 years (R=1250
mm/yr, K=50 m/d, and Qp = 10 l/s)
Figure 5 shows the relation between pumping rate
and saltwater intrusion length after 10 years with
227
return to menu
seawater boundary. The results show that a
higher pumping rate of 20 L/s near the coast
causes saltwater intrusion and contamination of
wells much quicker compared to a pumping rate
of 10 L/s. As expected, no pumping near the coast
resulted in slow movement of saltwater wedge in
the aquifer as shown in Figure 8. This indicates a
future direction for saltwater wedge management
in the lower Burdekin.
kappa
30
Pumping rate [L/s]
constant recharge rate of 1250 mm/yr and
hydraulic conductivities of 10, 50 and 100 m/d.
The results show that lower pumping and higher
recharge can reduce saltwater intrusion
effectively.
20
25
15
35000
Concentration (mg/L)
10
5
0
1000
K=10 m/d
2000
3000
4000
Intrusion length [m]
K=50 m/d
5000
K=100 m/d
Qp = 5 L/s
Qp = 15 L/s
30000
Qp = 10 L/s
Qp = 20 L/s
25000
20000
15000
10000
5000
0
Figure 5. Pumping rate vs. saltwater intrusion
length after 10 years (R=1250 mm/yr)
0
10
20
30
40
50
Time [years]
Figure 7. Evolution of salinity in the pumped
water from the well located 3000m from the
original saltwater boundary (1250 mm recharge,
K = 50 m/d)
For a non-homogeneous aquifer such as the
Burdekin
Delta
aquifer,
the
hydraulic
conductivity changes with the medium. Several
model runs were also carried out to study the
effects of varying hydraulic conductivity on salt
concentration profiles and intrusion lengths.
Preliminary simulations show that the system is
far more sensitive to recharge and pumping rates
than aquifer properties such as hydraulic
conductivity. Our simplified model developed in
this work has assumed homogeneous aquifer
properties. The effect of recharge on saltwater
intrusion length for a pumping rate of 10 L/s is
depicted in Figure 6. Higher hydraulic
conductivities cause broader dispersion zones (not
shown here) enhancing the total intrusion length.
Intrusion length is inversely proportional to
recharge rates.
35700
32130
28560
24990
21420
17850
14280
10710
7140
3570
1400
Recharge [mm/yr]
1200
K=10 m/d
K=50 m/d
K=100 m/d
1000
800
Figure 8. Salt concentration profiles (ppm)
without pumping near the coast after (A) 10
years, and (B) 50 years of simulation
600
400
200
Currently no bores are monitored along the model
transect, and in the area affected by saltwater
intrusion in the Burdekin Delta. A direct
comparison of the model results with the field
data is therefore not possible. For a typical run
with average aquifer parameters, the groundwater
salinity (t10 % of seawater salinity) moves about
3 km in the aquifer in 30 years. Comparison of
the modelling results has been made with limited
interpreted data of Arunakumaren et al. (2000)
and are found to be in reasonable agreement.
0
1000
1500
2000
2500
3000
3500
4000
4500
5000
Intrusion length [m]
Figure 6. Effect of recharge on saltwater
intrusion length after 10 years (Qp = 10 L/s)
Salinity in pumped waters along the model
transect were also analysed. Figure 7 illustrates
the change in pumping water salt concentration
with time for the well located at 3000m from the
228
return to menu
5.
intrusion of seawater, Proceeding: First
International Conference on Saltwater
Intrusion and Coastal Aquifers – Monitoring,
Modelling and Management, Essaouira,
Morocco, April 23-25 2001.
CONCLUSION
A 2-D variable-density groundwater flow and
solute transport model SUTRA has been used in
cross-section to simulate saltwater intrusion in the
Lower Burdekin Delta aquifer. The model
accounts for groundwater pumping, recharge rates
and hydrogeology of the aquifer system. The
results of this study highlight the effects of
pumping on saltwater intrusion. Less pumping
and high recharge rate in the aquifer can reduce
seawater intrusion effectively. Groundwater
pumping near the coast should be avoided to help
control the saltwater wedge. It is believed that
some of the results from this preliminary study
will be useful for water resources management in
the Lower Burdekin Delta.
Hillier, J. R. (1993) Management of saltwater
intrusion in coastal aquifers in Queensland,
Australia. AGSO Journal of Australian
Geology & Geophysics, 14 (2/3), 213-217.
Murphy, S. F., and R. C. Sorensen, Saltwater
intrusion in the Mackay coastal plains aquifer.
Proc. Aust. Sugar Cane Technol., 23, 70-76,
2002.
Narayan, K. A. and D. Armstrong, Simulation of
groundwater interception at Lake Ranfurly,
Victoria, incorporating variable density flow
and solute transport, J. Hydrology, 165, 161184, 1995.
Acknowledgements: This work is funded by
CSIRO Land and Water and Queensland
Department of Natural Resources and Mines. We
would like to thank Peter Gilbey, Sean Murphy
and Pushpa Onta for useful discussion.
Reilly, T. E., and A. S. Goodman, Quantitative
analysis of saltwater freshwater relationships
in groundwater systems – a historical
perspective, J. Hydrology, 80, 125-160, 1985.
6. REFERENCES
Arunakumaren, N. J., G. A. McMahon, and K.
Bajracharya, Water management in the lower
Burdekin:
Groundwater
model
conceptualization, Dept. of Natural Resources,
Brisbane, 2000.
Sinclair Knight Merz (1996) North and South
Burdekin Water Boards – Burdekin River
Issues. Sinclair Knight Merz.
Volker, R. E. and Rushton, K. R. (1982) An
assessment of the importance of some
parameters for seawater intrusion in aquifers
and a comparison of dispersive and sharpinterphase modelling approaches. J. Hydrol.
56, 239-250.
Charlesworth, P. B., K. A. Narayan, K. L.
Bristow, B. Lowis, G. Laidlow, and R.
McGowan, The Burdekin Delta – Australia’s
oldest recharge scheme, Proc. ISAR-4, 22-26
Sept., Adelaide, 2002.
Custodio, E., G. A. Bruggemen, and V.
Cotecchia, Groundwater problems in coastal
areas, Studies and Reports in Hydrology, vol.
35 , UNESCO, Paris 1987.
Voss, C.I. (1984) SUTRA: a finite element
simulation model for saturated-unsaturated,
fluid-density dependent groundwater flow
with energy transport or chemically reactive
single-species solute transport, US Geological
Survey, Water Resources Investigations,
Report 84-4369, pp 409.
Freeze, R. A., and J. A. Cherry, Groundwater,
Prentice Hall, New York, 1979.
Ghassemi F., A. J. Jakeman, and G. Jacobson,
Mathematical modelling of seawater intrusion,
Nauru Island, Hydrological Processes, 4, 269281, 1990.
Ghassemi, F., T. H. Chen, A. J. Jakeman, and G.
Jacobson, Two and three - dimensional
simulation
of
seawater
intrusion:
performances of the “SUTRA” and “HST3D”
models, AGSO J. Australian. Geology. &
Geophysics, 14(2/3), 219-226, 1993.
Gotovac, H., R. Andrecevic, and M. Vranjes,
Effects of aquifer heterogeneity on the
229
Effects Of Groundwater Pumping On Saltwater Intrusion
In The Lower Burdekin Delta, North Queensland
K. A. Narayana, C. Schleebergerb, P. B. Charleswortha and K. L. Bristowa
a
CSIRO Land and Water, Davies Laboratory, Townsville, Australia
b
University of Applied Sciences, Höxter, Germany
Abstract: The Burdekin Delta is situated in the dry tropics of North Queensland and provides a major water
resource for the irrigation of sugarcane. The Delta is unique in that it overlies a shallow groundwater system
and is close to the Great Barrier Reef. Aquifer management practices include large recharge pits to assist with
artificial replenishment of groundwater. Artificial recharge can be used to maintain groundwater levels and
subsequently control seawater intrusion. This technique, however, is often costly and ineffective in areas
where excessive groundwater pumping occurs. In the lower Burdekin Delta more than 1800 groundwater
pumps are used for irrigation purposes and excessive pumping has resulted in seawater intrusion near the
coastline. In this paper we describe the use of a variable density flow and solute transport model, SUTRA, to
define the current and potential extent of saltwater intrusion in the Burdekin Delta aquifer under various
pumping and recharge conditions. A 2-D vertical cross-section model has been developed for the area, which
accounts for groundwater pumping and various artificial recharge sites currently being used in the delta. The
Burdekin Delta aquifer consists mainly of sand and clay with granitic bedrock. The model domain uses
vertical cross sections along the direction of groundwater flow. The initial conditions were based on the land
use prior to agricultural development in the area when the seawater interface was in its assumed natural state.
The results address the effects of variations in pumping and artificial and natural recharge rates on the
dynamics of saltwater intrusion. The simulation has been carried out for a range of recharge, pumping rates
and hydraulic conductivity values. The results show that the saltwater intrusion is far more sensitive to
pumping rates and recharge than aquifer properties such as hydraulic conductivity. The impacts of possible
management scenarios on groundwater quality have also been investigated.
Keywords: Groundwater pumping, saltwater intrusion, SUTRA model, artificial recharge, Burdekin Delta
1.
Sorensen, 2001) and the first step in dealing with
these problems is to evaluate the size and extent
of the intrusion. The extent of this intrusion
depends on climatic conditions, the characteristics
of the groundwater flow within these aquifers,
and the manner of groundwater usage.
INTRODUCTION
The coastal areas of the world accommodate high
population with about 50% of the world
population lives within 60 km of the shoreline.
Overexploitation of the groundwater has become
a common issue along the coast where good
quality groundwater is available. Consequently,
many coastal regions in the world experience
extensive saltwater intrusion in aquifers resulting
in severe deterioration of the quality of
groundwater resources.
Saltwater intrusion problems in coastal aquifers
are not new. The initial model was developed
independently by Ghyben in 1888, and by
Herzberg in 1901. This simple model is known as
the Ghyben-Herzberg model and is based on the
hydrostatic balance between fresh and saline
water in a U-shaped tube. They showed that the
saltwater occurs at a depth h below sea level
represented by:
In Australia, coastal Queensland is fortunate to
have extensive groundwater resources. Many
rivers have well-developed alluvial tracts and
deltas with extensive sand and gravel aquifers.
The river delta systems usually contain rich soil
and were an obvious target for the development of
sugarcane plantations in the late 19th century.
Groundwater use for irrigation commenced
shortly after settlement, but it was not until the
expansion of the sugar industry in the mid-20th
century that irrigation was practiced extensively.
Serious problems of saltwater intrusion exist in
many coastal areas of Queensland (Volker and
Rushton, 1982; Hillier, 1993; Murphy and
h = >Us (Us-Uf)@ hf
(1)
where, Uf and Us are respectively the density of
fresh and saline water, and hf is the elevation of
fresh water level above mean sea level. Equation
1 is referred to as Ghyben-Herzberg relationship.
Substitution of Uf (1000 kg m-3) and Us (1025 kg
m-3) in eqn. (1) shows that h = 40 hf. In other
words, the depth to the fresh-saline interface
below mean sea level (h) is 40 times the elevation
224
return to menu
sugarcane areas in Australia. The topography is
essentially flat to slightly undulating, however,
outcrops of basement rock occur in the south and
southwest of the delta. Annual rainfall for the
region averages 1032 mm with most falling in the
December – March ‘wet’ season. Mean annual
evaporation is 2062 mm. The delta is situated in
close proximity to environmentally sensitive
wetlands, waterways, estuaries, and the Great
Barrier Reef. The location of the delta with regard
to its regional setting is shown in Figure 1.
of the water table above sea level (hf) (Freeze and
Cherry, 1979).
This simplistic model ignores convection,
dispersion and diffusion phenomena responsible
for the distribution of salinity in coastal aquifers.
In coastal aquifers, freshwater usually overlies the
seawater separated by a transition zone.
Management of limited groundwater resources in
such situations is a delicate task and requires
special attention to minimise the movement of the
saltwater wedge into aquifers and upconing of
saltwater near pumping stations. The extent of
intrusion depends on a number of factors such as
aquifer geometry and properties (hydraulic
conductivity,
anisotropy,
porosity
and
dispersivity), abstraction rates, and depth,
recharge rate, and distance of pumping wells from
the coastline (Ghassemi et al, 1993). Complex
models are required to quantify these factors.
The history of groundwater irrigation, aquifer
overdraft and consequent formation of the
Burdekin Delta recharge scheme is summarised
by Charlesworth et al. (2002). One of the main
purposes of the lower Burdekin artificial recharge
scheme is to maintain high water table levels to
keep the saltwater wedge from moving inland and
displacing the freshwater. Currently there are
more than 1800 groundwater pumps used for
irrigation water supply in the delta area. It is
critical to understand the effectiveness of the
artificial recharge scheme to withstand the
movement of saltwater wedge under various
irrigation and pumping regimes.
Over the years, several mathematical and
numerical models have been developed, which
serve to predict the interface or transition zone
between fresh groundwater of meteoric origin and
seawater in the subsurface of coastal areas (Reilly
and Goodman, 1985). The development of these
models was largely motivated by groundwater
issues; that is, assessment of fresh groundwater
reserves, and prediction of saltwater intrusion
(onshore salinity distribution) – the landward or
upward movement of the interface in response to
groundwater exploitation practices (e.g. Volker
and Rushton, 1982; Custodio et al., 1987;
Ghassemi et al., 1990; Ghassemi et al., 1993;
Gotovac et al., 2001).
In this paper we describe the use of a variable
density flow and solute transport model, SUTRA
(Voss, 1984), to define the current and potential
extent of saltwater intrusion in the Burdekin Delta
aquifer under various pumping and recharge
conditions. A 2-D vertical cross-section model
has been developed for the area, which accounts
for groundwater pumping and various artificial
recharge schemes currently being used in the
delta. Model results are compared with limited
available data in the area. The impacts of possible
management scenarios on groundwater quality
have also been investigated.
2.
Figure 1. Location of the Lower Burdekin Delta
(BRIA - Burdekin River Irrigation Area, NBWB
– North Burdekin Water Board and SBWB –
South Burdekin Water Board)
A single-layer approach to modelling the
Burdekin Delta aquifers allows all minor
variations to be incorporated into a single
hydrostratigraphic unit (Arunakumaren et al.,
2000). Groundwater within the delta is
constrained by the unconsolidated alluvial and
deltaic sediments, which were deposited by the
Burdekin River and its distributaries. Typical
cross-sections of the delta sediments show that
strata layers usually are laterally discontinuous,
making cross-bore correlations difficult to
identify. Nonetheless, it is apparent from bore
hydrographs that there is a significant amount of
vertical hydraulic connection between sandy
units.
SITE DESCRIPTION AND GENERAL
HYDRGEOLOGY
The Burdekin Delta is a major irrigation area in
the dry tropics of North Queensland, with about
40,000 ha of irrigated crops. The environmental
and climatic conditions are ideal for the
production of sugarcane and the region has a
reputation of being one of the highest yielding
225
return to menu
require adequate discretisation in space and time
and appropriate initial and boundary conditions.
In this work the delta is being simulated as a
single-layer unconfined aquifer, with a bottom
slope defined by the position of the granitic
bedrock. The hydraulic conductivities of the
aquifer have been estimated based on aquifer
materials (Freeze and Cherry, 1979). Ongoing
improvements in hydraulic characterization will
be made as more data becomes available. A
typical cross section of the delta aquifer is shown
in Figure 2.
3.
3.2 SUTRA model
SUTRA (Voss, 1984), in conjunction with the
Argus-One Graphic User Interface, was chosen as
the basis for numerical modelling because of its
ability to solve density dependent groundwater
flow and variably saturated flow, and also
because it is readily available in source code
form. This model implements a hybridisation of
finite element and integrated finite difference
methods employed in the framework of a method
of weighted residuals. The hybrid method is the
simplest and most economical approach, which
preserves the mathematical elegance and
geometric flexibility of finite element simulation,
while taking advantage of finite difference
efficiency.
NUMERICAL MODELLING OF
SALTWATER INTRUSION
3.1 Recharge and groundwater extraction
Recharge to the groundwater system is by a
number of different processes. These include
infiltration of rainfall, artificial recharge through
pits and channels, river recharge, flooding, and
irrigation return flows. From an assessment of
these mechanisms, simulated annual groundwater
recharge between 1981 and 1997 varied between
330,000
and
650,000
ML
per
year
(Arunakumaren et al., 2000). For analyses in this
study, total maximum recharge rate was set at
1,250 mm/yr (490,000 ML/yr.
3.3 Discretisation, Boundary Conditions and
Model Parameters
To simulate saltwater intrusion in the Burdekin
Delta aquifer, a cross-sectional vertical slice with
a length of 5000 m, a depth of 30 m on the left
boundary, 45 m on the right boundary (facing
saltwater), and a thickness of 1 m was taken. The
model area was discretised to 7500 (500 x 15)
quadrilateral elements and 8016 (501 x 16) nodes.
The details of the model domain with boundary
conditions are shown in Figure 3.
Simulation of the saltwater intrusion requires a
series of assumptions and estimations to be made
about the various parameters involved. There are
no volumetric metered records for groundwater
pumping rates in the Burdekin Delta. The most
recent estimate indicates a volume of 210,000530,000 ML/yr is being extracted (Arunakumaren
et al., 2000; SKM, 1996).
Figure 2. A cross-section of the Burdekin Delta in
groundwater flow direction showing sediment
distribution (after Arunakumaren et al. 2000)
The physical problem along the coast in the Delta
is one of density-dependent groundwater flow and
saltwater intrusion (Voss, 1984; Narayan and
Armstrong, 1995). Prediction of the saltwater
wedge can be obtained by the solution of two
spatial differential equations describing the
‘conservation of mass of fluid’ and ‘conservation
of mass of salt’ in a porous medium. Numerical
procedures for solving the governing equations
226
Figure 3. Model domain and boundary
conditions
A steady state simulation with 200 mm natural
recharge per year was used to generate the
situation prior to agricultural development. The
result is a saltwater wedge that extends about 50
m inland. Concentration and pressure outputs
from steady state SUTRA runs were used as an
initial condition to simulate realistic short-time
effects of agricultural land use (groundwater
discharge and recharge) on the groundwater
system.
return to menu
The boundary conditions for the transport
simulation are dependent on its flow boundary
conditions. On the right hand model boundary a
hydrostatic pressure was imposed where pressure
is zero at the sea surface, and increases linearly
with depth. Based on field observations, a
groundwater head of 2 m was chosen at a distance
of 5km from the right hand model boundary. The
concentration (TDS) of the recharge water is
assumed to be 300 mg/L owing to irrigation with
river- and groundwater.
Longitudinal dispersivity
2.5 m
Transverse dispersivity
0.5 m
Molecular diffusivity of solute in
water, Dm
10-9 m²/s
Porosity, İ
0.3
Viscosity of water, P
0.001 kg·m/s
Groundwater head (left of model
boundary)
2m
Recharge range
500 – 1250 mm/yr
The hydraulic conductivity varies across the delta
from 10 to >300 m/d at the northern coast. In the
area chosen for the SUTRA model (see Figure 2)
K is estimated to be 10 to 100 m/d (Freeze and
Cherry, 1979), salinity of groundwater in the
cross-section as 500 mg/L (entire Burdekin Delta
aquifer ranges: 300 to 2000 mg/L). Transverse
dispersivity is even less well quantified than
longitudinal dispersivity from field observations.
The dispersivity values listed in Table 1 are an
estimate only, governed to some extent by the
practical limitations imposed by mesh size.
Distance between wells, Lp
400 m
Pumping rate, Qp
5 – 50 L/s
4.
One of the main purposes of the lower Burdekin
artificial recharge scheme is to maintain high
water table levels to keep the saltwater wedge
from moving inland and displacing the
freshwater. It is important to understand the
effects of varying recharge and groundwater
pumping on the movement of the saltwater
wedge. Based on the work of Arunakumaren et al.
(2000), simulation runs were conducted with
recharge values of 600, 900 and 1250 mm/yr. The
actual recharge differs seasonally during a year.
SUTRA cannot simulate the seasonal variations
properly, so the constant source value for
recharge is spread over one year (e.g. 1250 mm/yr
§ 0.0000396 kg/(m²·s)).
A total of 2187 pump locations were surveyed
across the delta by GPS, including 367 open
water pumps and 1811 groundwater pumps by
NR&M (Arunakumaren et al., 2000). In this
model, 12 perforated wells, with discharge rates
of 5, 10, 15, 20, 25 or 50 L/s and a depth of about
10 m have been used along the transect. Nodes
with negative fluid source describe these wells.
Average calculated distance between pumps Lp
[m] in the delta is about 470 m; for modelling
purposes we used Lp = 400 m. The groundwater
pumping rates vary across the delta because of
different irrigation practices, field size, and
precipitation as well as surface water availability.
To obtain the linear discharge values for a onemeter thick vertical slice, the pumping rates were
divided by Lp. For example, 10 L/s pumping rate
can be converted to a line sink (Qp') by 10/400
(kg/m·s) = 0.025 kg/(m·s). Model input
parameters are given in Table 1.
The pumping rates used in the model are 5, 10,
15, 20, 25 and 50 L/s. This corresponds to 0.4324.32 ML/day. The first four values are more
realistic for the Burdekin Delta. A series of salt
concentration contours for various recharge rates
(600-1250 mm/yr), hydraulic conductivities (10 100 m/d) and pumping rates (5–50 L/s) were
obtained. Saltwater concentration profiles for
1250 mm/yr recharge rate and a typical pumping
rate of 10 L/s are shown in Figure 4 after 10
years. Concentration profiles are influenced by
the left hand boundary for longer pumping
durations.
Table 1. Input parameters.
Value
Seawater concentration, CS
35,700 mg/L
Groundwater concentration, CG
500 mg/L
Recharge water concentration, CR
300 mg/L
Freshwater density, ȡF
1,000 kg/m³
Density change, wȡ/wC
700 kg²/(kgTDS·m³)
Hydraulic conductivity, K
10, 50, and 100 m/day
Metres
Parameter
RESULTS AND DISCUSSION
Figure 4. Saltwater concentration profile in a
vertical cross-section after 10 years (R=1250
mm/yr, K=50 m/d, and Qp = 10 l/s)
Figure 5 shows the relation between pumping rate
and saltwater intrusion length after 10 years with
227
return to menu
seawater boundary. The results show that a
higher pumping rate of 20 L/s near the coast
causes saltwater intrusion and contamination of
wells much quicker compared to a pumping rate
of 10 L/s. As expected, no pumping near the coast
resulted in slow movement of saltwater wedge in
the aquifer as shown in Figure 8. This indicates a
future direction for saltwater wedge management
in the lower Burdekin.
kappa
30
Pumping rate [L/s]
constant recharge rate of 1250 mm/yr and
hydraulic conductivities of 10, 50 and 100 m/d.
The results show that lower pumping and higher
recharge can reduce saltwater intrusion
effectively.
20
25
15
35000
Concentration (mg/L)
10
5
0
1000
K=10 m/d
2000
3000
4000
Intrusion length [m]
K=50 m/d
5000
K=100 m/d
Qp = 5 L/s
Qp = 15 L/s
30000
Qp = 10 L/s
Qp = 20 L/s
25000
20000
15000
10000
5000
0
Figure 5. Pumping rate vs. saltwater intrusion
length after 10 years (R=1250 mm/yr)
0
10
20
30
40
50
Time [years]
Figure 7. Evolution of salinity in the pumped
water from the well located 3000m from the
original saltwater boundary (1250 mm recharge,
K = 50 m/d)
For a non-homogeneous aquifer such as the
Burdekin
Delta
aquifer,
the
hydraulic
conductivity changes with the medium. Several
model runs were also carried out to study the
effects of varying hydraulic conductivity on salt
concentration profiles and intrusion lengths.
Preliminary simulations show that the system is
far more sensitive to recharge and pumping rates
than aquifer properties such as hydraulic
conductivity. Our simplified model developed in
this work has assumed homogeneous aquifer
properties. The effect of recharge on saltwater
intrusion length for a pumping rate of 10 L/s is
depicted in Figure 6. Higher hydraulic
conductivities cause broader dispersion zones (not
shown here) enhancing the total intrusion length.
Intrusion length is inversely proportional to
recharge rates.
35700
32130
28560
24990
21420
17850
14280
10710
7140
3570
1400
Recharge [mm/yr]
1200
K=10 m/d
K=50 m/d
K=100 m/d
1000
800
Figure 8. Salt concentration profiles (ppm)
without pumping near the coast after (A) 10
years, and (B) 50 years of simulation
600
400
200
Currently no bores are monitored along the model
transect, and in the area affected by saltwater
intrusion in the Burdekin Delta. A direct
comparison of the model results with the field
data is therefore not possible. For a typical run
with average aquifer parameters, the groundwater
salinity (t10 % of seawater salinity) moves about
3 km in the aquifer in 30 years. Comparison of
the modelling results has been made with limited
interpreted data of Arunakumaren et al. (2000)
and are found to be in reasonable agreement.
0
1000
1500
2000
2500
3000
3500
4000
4500
5000
Intrusion length [m]
Figure 6. Effect of recharge on saltwater
intrusion length after 10 years (Qp = 10 L/s)
Salinity in pumped waters along the model
transect were also analysed. Figure 7 illustrates
the change in pumping water salt concentration
with time for the well located at 3000m from the
228
return to menu
5.
intrusion of seawater, Proceeding: First
International Conference on Saltwater
Intrusion and Coastal Aquifers – Monitoring,
Modelling and Management, Essaouira,
Morocco, April 23-25 2001.
CONCLUSION
A 2-D variable-density groundwater flow and
solute transport model SUTRA has been used in
cross-section to simulate saltwater intrusion in the
Lower Burdekin Delta aquifer. The model
accounts for groundwater pumping, recharge rates
and hydrogeology of the aquifer system. The
results of this study highlight the effects of
pumping on saltwater intrusion. Less pumping
and high recharge rate in the aquifer can reduce
seawater intrusion effectively. Groundwater
pumping near the coast should be avoided to help
control the saltwater wedge. It is believed that
some of the results from this preliminary study
will be useful for water resources management in
the Lower Burdekin Delta.
Hillier, J. R. (1993) Management of saltwater
intrusion in coastal aquifers in Queensland,
Australia. AGSO Journal of Australian
Geology & Geophysics, 14 (2/3), 213-217.
Murphy, S. F., and R. C. Sorensen, Saltwater
intrusion in the Mackay coastal plains aquifer.
Proc. Aust. Sugar Cane Technol., 23, 70-76,
2002.
Narayan, K. A. and D. Armstrong, Simulation of
groundwater interception at Lake Ranfurly,
Victoria, incorporating variable density flow
and solute transport, J. Hydrology, 165, 161184, 1995.
Acknowledgements: This work is funded by
CSIRO Land and Water and Queensland
Department of Natural Resources and Mines. We
would like to thank Peter Gilbey, Sean Murphy
and Pushpa Onta for useful discussion.
Reilly, T. E., and A. S. Goodman, Quantitative
analysis of saltwater freshwater relationships
in groundwater systems – a historical
perspective, J. Hydrology, 80, 125-160, 1985.
6. REFERENCES
Arunakumaren, N. J., G. A. McMahon, and K.
Bajracharya, Water management in the lower
Burdekin:
Groundwater
model
conceptualization, Dept. of Natural Resources,
Brisbane, 2000.
Sinclair Knight Merz (1996) North and South
Burdekin Water Boards – Burdekin River
Issues. Sinclair Knight Merz.
Volker, R. E. and Rushton, K. R. (1982) An
assessment of the importance of some
parameters for seawater intrusion in aquifers
and a comparison of dispersive and sharpinterphase modelling approaches. J. Hydrol.
56, 239-250.
Charlesworth, P. B., K. A. Narayan, K. L.
Bristow, B. Lowis, G. Laidlow, and R.
McGowan, The Burdekin Delta – Australia’s
oldest recharge scheme, Proc. ISAR-4, 22-26
Sept., Adelaide, 2002.
Custodio, E., G. A. Bruggemen, and V.
Cotecchia, Groundwater problems in coastal
areas, Studies and Reports in Hydrology, vol.
35 , UNESCO, Paris 1987.
Voss, C.I. (1984) SUTRA: a finite element
simulation model for saturated-unsaturated,
fluid-density dependent groundwater flow
with energy transport or chemically reactive
single-species solute transport, US Geological
Survey, Water Resources Investigations,
Report 84-4369, pp 409.
Freeze, R. A., and J. A. Cherry, Groundwater,
Prentice Hall, New York, 1979.
Ghassemi F., A. J. Jakeman, and G. Jacobson,
Mathematical modelling of seawater intrusion,
Nauru Island, Hydrological Processes, 4, 269281, 1990.
Ghassemi, F., T. H. Chen, A. J. Jakeman, and G.
Jacobson, Two and three - dimensional
simulation
of
seawater
intrusion:
performances of the “SUTRA” and “HST3D”
models, AGSO J. Australian. Geology. &
Geophysics, 14(2/3), 219-226, 1993.
Gotovac, H., R. Andrecevic, and M. Vranjes,
Effects of aquifer heterogeneity on the
229
